Fetching from the wire…
Public story · 2026-09-09 · high
Every example follows one manifest format that bundles a Codex skill, an app and an MCP server into a single install folder.
Why now: The repo's manifest format is the reference point developers are already building against, star count and all.
OpenAI's plugins repository has passed 6,072 stars, and the whole thing runs on one file format. Each plugin lives in plugins/<name>/ with a required .codex-plugin/plugin.json manifest, plus optional skills/, .app.json and .mcp.json files alongside it.
That folder structure is the documentation. There's no separate spec page explaining how Codex plugins are supposed to work. You read the manifest layout in a working example instead.
The format itself answers a real question: how do you package a Codex skill, an app and an MCP server connection so they install as one unit? The manifest is the answer. A plugin folder can carry all three, wired together through plugin.json.
For anyone building on Codex, that's the reference to copy now rather than guess at later. The repo is curated, not a dump of every plugin in the wild. What's inside reads closer to a template than a sample of real-world usage.
What the repo doesn't say is how OpenAI validates a plugin before anyone trusts it, or whether this manifest shape holds as Codex changes. Watch for a formal spec document or an SDK release. Either would mean OpenAI is locking the format in writing, not just letting a popular repo define it by example.
Each link below shares sources, entities, or timing with this story.
A GitHub repo cataloging Claude Code tips doesn't normally warrant a top story. But shanraisshan/claude-code-best-practice at 53.4K stars isn't a tips list anymore. It's the de facto reference for how an entire generation of developers is learning to work with AI coding agents...
The result-interception hook (#41202) is a place to redact secrets or strip injected instructions out of untrusted MCP output before it enters context, which is the missing seam in every MCP setup I've built. Also: a configurable grace period for discovering tools from optiona...
A Chinese lab shipped a runtime that manages two American coding agents as subagents, and it went from repo creation to 145,439 stars in four days. deepseek-ai/deepseek-harness published dsh-v0.1.0-rc.7 at 12:01 UTC today, its first tagged release since the repo appeared on Au...
Three things happened this month that only make sense together. Agent Plugins 1.0 shipped co-signed by six competitors: AWS, Anysphere, Microsoft, OpenAI, Vercel and Google (GitHub Changelog). It makes skills-plus-MCP bundles portable across clients. OpenAI's August 11 Codex c...
Six clients. One manifest. Zero vendor lock. Vercel published Agent Plugins 1.0.0 on August 6, an openly licensed spec that bundles Agent Skills and MCP servers behind a single portable manifest. The shape is deliberately boring: a plugin.json requiring only schemaVersion and...
Language packs like samber/cc-skills-golang, 1,600-plus cross-tool libraries, harnesses like DeerFlow with built-in skills, even Google's Workspace CLI bundling agent skills. The unit of reuse shifted from prompts and MCP tools to portable markdown skills that travel across Cl...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.